Getting a model to return JSON is easy.
Getting reliable JSON across multiple models is a production problem.
A response can parse successfully and still break the workflow:
a required field is missing
an enum value is unsupported
Getting a model to return JSON is easy. Getting reliable JSON across multiple models is a production...
Getting a model to return JSON is easy.
Getting reliable JSON across multiple models is a production problem.
A response can parse successfully and still break the workflow:
a required field is missing
an enum value is unsupported

Getting an AI API request to return a response is only the beginning. For real AI products, the...

Your LLM Can Return Perfect JSON and Still Be Completely Wrong Most developers (including...

A 1.5B model running on your laptop will return JSON that almost parses. The closing brace is...

Every developer who has worked with LLMs has been there. You ask the model for JSON. You describe the...

For a long time the trick to force a model into JSON was to prefill the assistant turn with an...

The layer between 'the model usually returns the right shape' and 'the rest of my system can depend on it' — schemas, validation…